Backend Requirements

Specification for the ASP.NET / Azure backend powering this frontend prototype

Purpose

This frontend demonstrates the eight modules described in the RFP using mock data. The list below consolidates every backend capability the UI consumes, mapped to the RFP sections (5–11) and to a recommended Azure / ASP.NET service. Use it as the input to your API contract design and Azure landing-zone planning.

Architecture Diagram (Interactive)

INTEGRATION LAYER
Cloud Integration
Microsoft 365 /
Teams /
SharePoint
Power BI
External Portals
GeBIZ Tender Portal
SAP Ariba
Other External Tender Portals
Future APIs
(Extensibility)
L1: USER INTERFACE (UI) & PORTAL
User Prompt | Configure Agents | Monitor Status | View Reports
L2: API GATEWAY
API Gateway Management
L3: QUEUE MANAGEMENT
Agent Job Queue
L4: EXECUTION
MEMORYState /
Context Store
1. TENDER INTELLIGENCE
Search & analysis of global tender opportunities, RFP parsing & Requirement extraction, Competitor Insight.
2. GO/NO-GO ADVISOR
RFP Fit analysis. Win probability. PBRA Risk indicator. Resource alignment.
3. RESOURCING & SUBCONSULTANT MATCHER
Internal staff availability. Skills mapping. Subconsultant database & vetting.
4. BID STRATEGY & PROPOSAL DRAFTING
Bid Strategy recommendation, win-theme development, proposal section drafting, and RAG-grounded content generation
5. COMPLIANCE REVIEW
submission readiness checker, gap analysis against RFP requirements
Validated Execution Plan & Bid Data
LLM & MODELSModel Registry /
Router
L5: BID GOVERNANCE SERVICES
Workflow for Approvals
& Submission
Governance
Controls
Analytics
Bid Data Lake & AI Model Ownership
L6: AI GUARDRAILS AND GOVERNANCE
PII Masking
Prompt Injection
Guardrail
Data Privacy
Filter
Content Safety
Filter
Human Override Required
Governance Reports
L8: AI MODEL GOVERNANCE & ROUTING
L7: OUTPUT & REPORTS
Validated Tender Documents
Final Audited Output
L9: OBSERVABILITY
Management Control
System Monitoring
(Traces + Metrics)
Admin Console
L10. SECURITY
(Cross-cutting)
Data
Classification
ID
Propagation
Secrets
Manager

1. Recommended Azure / ASP.NET Stack (L1-L3)

API Layer (L2)
Azure API Gateway + ASP.NET Core 8 Web API
Queue Management (L3)
Azure Service Bus / Event Hubs for Agent Job Queue
Auth & Security (L10)
Microsoft Entra ID — OIDC, RBAC, ID Propagation
Primary DB
Azure SQL Database (hyperscale) + EF Core
Document store
Azure Blob Storage (private, CMK encryption)
Search & RAG
Azure AI Search (hybrid: vector + BM25) + pgvector on Postgres Flexible (optional)
Workflow engine
Azure Durable Functions or Elsa Workflows for approvals
Integration
Azure API Management + Logic Apps for GeBIZ, Ariba, M365, PMS, HR/ERP
Reporting
Power BI Embedded (REST API + workspace per tenant)
Hosting
Azure Container Apps or App Service (Linux), AKS optional for scale
Observability
Application Insights + Log Analytics + Defender for Cloud
Region
Southeast Asia (Singapore) — satisfies §9.6 data residency

2. Core Data Model (entities)

Opportunity
Source (GeBIZ/Ariba/Manual), client, sector, value, dates, status, AI match score, tags
Bid
Code, title, lead partner, team, value, status, Go/No-Go, risk level, win probability, submission date
BidTeamMember
Bid ↔ User with role & allocation %
Subconsultant
Profile, capability tags, historical engagements, rating
Document
Blob URI, version, type, checksum, retention class
ApprovalWorkflow / Step / Action
Configurable matrix, SLA, escalation, full audit history
RiskItem
Title, severity, mitigation, owner, linked clauses
KnowledgeArticle
Category, content, embedding ref, usage count, tags
ProjectTrackRecord
Past project metadata used as references in proposals
AuditLog
Append-only — user, action, entity, before/after, timestamp, IP
AIInteraction
Prompt, model, tokens, cost, RAG sources, user, response, evaluation
Notification
User, channel (in-app/email/Teams), payload, read state
MarketIntelMetric
Competitor analysis, win/loss rates, margin estimates, market share
AgentWorkflow
AI multi-agent pipelines, step definitions, execution triggers
SimulationRun
Execution history of agent workflows, trace logs, success/fail metrics
AnalyticsReport
Aggregated system-wide data cubes for pipeline, sector distributions, revenue
SystemHealth
API latency, uptime, CPU/Memory telemetry (usually backed by App Insights)
LLMOpsLog
Cost-per-token, drift alerts, latency, fallback events for Admin console

3. REST API Surface (representative)

MethodEndpointPurposeRFP §
Opportunities & Bids
GET/api/opportunitiesList + filter live tender feed5.2.1
POST/api/opportunities/sync/{source}Trigger pull from GeBIZ / Ariba5.2.1, 7
POST/api/bidsRegister a bid (auto or manual)5.2.2
GET/api/bids?status=&sector=Filterable bid registry5.2.2
GET/api/bids/{id}Bid detail with team, docs, risks5.2.3
POST/api/bids/{id}/teamAllocate resources & subconsultants5.2.3
External Partners & Sourcing
GET/POST/api/partnersList, search & pre-qualify specialist partners & suppliers5.2.3, 5.2.5
GET/api/partners/{id}/scorecardMulti-metric partner performance history & CVs5.2.3, 5.2.8
POST/api/rfq-packagesCreate RFQ work packages & invite consultants5.2.3, 5.2.5
GET/POST/api/quotations/compareSide-by-side quotation leveling & award sign-off5.2.3, 5.2.6
GET/api/pricing/benchmarksHistorical discipline unit rates & inflation indexes5.2.3, 5.2.8
Workflow & Submissions
POST/api/approvals/{bidId}/submitKick off approval workflow5.2.6
POST/api/approvals/{stepId}/decisionApprove / reject / escalate5.2.6
POST/api/submissions/{bidId}/checklistCompliance checklist validation5.2.7
POST/api/submissions/{bidId}/finalizeLock & submit5.2.7
AI Intelligence & Copilot
POST/api/ai/summariseSummarise tender document5.2.4, 8
POST/api/ai/risk-scanDetect onerous clauses5.2.4, 8
POST/api/ai/go-no-goScore Go/No-Go with rationale5.2.4
POST/api/ai/recommend-teamSuggest team & subconsultants5.2.4
GET/api/ai/workflowsList custom multi-agent workflows8
POST/api/ai/workflows/{id}/simulateExecute workflow simulation8
POST/api/proposals/{bidId}/draftAI-assisted proposal draft5.2.5
Knowledge Hub
POST/api/knowledge/searchHybrid RAG search5.2.5, 8.2.3
GET/PUT/api/documents/{id}Versioned document mgmt (Blob-backed)5.2.5
GET/api/knowledge/assetsList proposal assets, lessons learnt5.2.5
Analytics & Market Intel
POST/api/results/{bidId}Record win/loss + reasons5.2.8
GET/api/analytics/win-lossAggregated reporting5.2.8
GET/api/market-intel/competitorsCompetitor margin & overlap frequency5.2.8
GET/api/analytics/pipelineRevenue projections & sector performance5.2.8
Platform Administration
GET/api/audit?entity=&from=&to=Audit log query (admin)9.2
GET/api/admin/healthSystem health & API latency metrics10, 11
GET/api/admin/llm-opsToken usage, cost, model drift alerts8, 9
POST/api/admin/rolesRBAC management & policy sync9.1

4. Integration Layer (Cloud & External Portals)

GeBIZ Tender Portal
Scheduled crawl/API ingestion → Service Bus → Opportunity normaliser
SAP Ariba
OAuth API + delta sync; map procurement events → Opportunity
Microsoft 365 / Teams
Graph API for files, channels, meetings, notifications
SharePoint
Optional document mirror; Graph API + change feed
Power BI
Embedded reports via Power BI REST; row-level security per role
Other External Portals
REST/SOAP through APIM façade with retry + circuit breaker
Future APIs
Versioned public API + webhook framework for extensibility

5. AI Execution & Guardrails (L4, L6, L8)

LLM & Models (L4/L8)
Azure OpenAI Model Registry & Router (GPT-4o primary, Phi-3 fallback)
Embeddings
text-embedding-3-large; chunked at 800 tokens with 100 overlap
Memory (State/Context)
Azure AI Search vector index per tenant + metadata filters
Orchestration (L4)
Semantic Kernel agents (Tender Intel, Go/No-Go, Resourcing, Drafting, Compliance)
RAG pipeline
Doc ingest (Form Recognizer) → chunk → embed → index → retrieve → rerank → ground
AI Guardrails (L6)
PII Masking, Prompt Injection Guardrail, Data Privacy Filter, Content Safety Filter
Governance Reports
Prompt + response logging, evaluation harness, drift alerts (requires Human Override optionally)
Cost control
Token budget per user/day, caching layer, response streaming

6. Bid Governance Services (L5 & L7)

  • · Configurable Workflow for Approvals & Submission by bid value, sector, risk level (admin-editable)
  • · Multi-level Governance Controls with parallel & sequential steps, delegations and out-of-office
  • · Bid Data Lake & AI Model Ownership: tracks telemetry for Analytics and AI models
  • · Validated Tender Documents and Final Audited Output generated at L7 for Management Control
  • · SLA tracking with automatic escalation to next-level approver
  • · Full audit trail (append-only) covering every decision, comment and attachment
  • · In-app + email + Teams notifications via Azure Communication Services + Graph
  • · Recommendation: Azure Durable Functions for long-running orchestrations

7. Security (L10 Cross-cutting)

ID Propagation
OAuth On-Behalf-Of (OBO) flow for end-to-end identity propagation
Data Classification
Microsoft Purview for auto-tagging and data classification
Secrets Manager
Azure Key Vault, managed identities everywhere — no static secrets
SSO + MFA
Entra ID (Azure AD) with Conditional Access policies
RBAC
App roles: Admin, Partner, BidLead, Contributor, Approver, Viewer
Encryption
TLS 1.3 in transit; AES-256 at rest with customer-managed keys
Audit
Immutable Append Blob audit store + Azure Monitor; 7-year retention
Backups
PITR for SQL (35 days), geo-redundant blob backups, monthly DR drills
DR
Active-passive across SEA & East Asia; RTO 4h / RPO 1h target
Data residency
Singapore primary; replicas only within ASEAN regions
Threat protection
Defender for Cloud, WAF on Front Door, DDoS Standard

8. Observability & Non-Functional (L9)

System Monitoring
App Insights traces + log analytics (L9 Traces + Metrics)
Admin Console
Unified L9 observability dashboard (implemented in IT Admin Console)
Availability
99.9% monthly — multi-AZ deployment, health probes, blue/green releases
Concurrent users
20–40 baseline; scale to 200 burst (Container Apps autoscale)
Bid volume
500–600 bids/year, 5–30 docs each — sized for ~3 TB year 1
AI throughput
~300 queries/day with PTU + on-demand fallback
Search latency
p95 < 800 ms for list/search; p95 < 3 s for RAG response
Document upload
Resumable up to 500 MB via Blob SAS URLs
Observability
App Insights traces + custom KPIs (win rate, AI cost/bid, approval SLA)

9. DevOps & Delivery

Source control
Azure DevOps Repos / GitHub Enterprise
CI/CD
GitHub Actions or ADO pipelines → Bicep/Terraform IaC → Container Apps
Environments
Dev / SIT / UAT / Prod with isolated subscriptions
Testing
xUnit + Playwright + k6 load + Pact contract tests for integrations
Quality gates
SonarCloud, OWASP ZAP, Defender for DevOps in PR pipeline
Suggested next steps
  1. Confirm the API contract for each endpoint above (OpenAPI spec).
  2. Stand up the Azure landing zone (Entra ID, subscriptions, networking, Key Vault).
  3. Implement the Opportunity ingestion + Bid registry first — they unlock the rest of the UI.
  4. Wire the AI module to Azure OpenAI behind a feature flag for controlled rollout.
  5. Replace the mock data in src/lib/mock-data.ts with live API calls (TanStack Query).

AI Bid Assistant

Pursuite

Hello! I'm your Pursuite AI Assistant. How can I help you with your bid management today?

11:17 AM
Suggested for this page